Improve arithmetic operations involving CFTimeIndexes and TimedeltaIndexes

[spencerkclark]

• Closes Enable add/sub operations involving a CFTimeIndex and a TimedeltaIndex #2484
• Tests added (for all bug fixes or enhancements)
• Fully documented, including whats-new.rst for all changes and api.rst for new API

[pep8speaks]

Hello @spencerkclark! Thanks for submitting the PR.

• There are no PEP8 issues in the file xarray/coding/cftimeindex.py !

• There are no PEP8 issues in the file xarray/tests/test_cftimeindex.py !

[shoyer]

Has this been changed now? As you know, it's slightly dangerous to dip into pandas internals...

[spencerkclark]

Yes, this method is not needed with recent versions of pandas. I think the change that eliminated the need for it was in this PR: pandas-dev/pandas#19895, which I guess would have made it into pandas version 0.23.0.

I added it because of the failing test in the Windows 32-bit Python 2.7 build that used pandas version 0.22.0. Should I just mark the test as an expected failure for versions of pandas less than 0.23 and remove this private method?

[shoyer]

Just add a brief comment here summarizing this, e.g., "# no longer used as of pandas 0.23"

…

On Mon, Oct 15, 2018 at 5:36 AM Spencer Clark ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In xarray/coding/cftimeindex.py <#2485 (comment)>: > return CFTimeIndex(other + np.array(self)) def __sub__(self, other): - return CFTimeIndex(np.array(self) - other) + if isinstance(other, CFTimeIndex): + return pd.TimedeltaIndex(np.array(self) - np.array(other)) + elif isinstance(other, pd.TimedeltaIndex): + return CFTimeIndex(np.array(self) - other.to_pytimedelta()) + else: + return CFTimeIndex(np.array(self) - other) + + def _add_delta(self, deltas): + # To support TimedeltaIndex + CFTimeIndex with older pandas versions Yes, this method is not needed with recent versions of pandas. I think the change that eliminated the need for it was in this PR: pandas-dev/pandas#19895 <pandas-dev/pandas#19895>, which I guess would have made it into pandas version 0.23.0. I added it because of the failing test in the Windows 32-bit Python 2.7 build that used pandas version 0.22.0. Should I just mark the test as an expected failure for versions of pandas less than 0.23 and remove this private method? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2485 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABKS1iKDRM5Cgp8ODnM18-bYY1rqidWmks5ulH9dgaJpZM4Xaojb> .